@boldceo@bcherny@claudeai for reference... this is a typical session. i've been working a few projects simultaneously since last night (with a break for sleep of course!)
@boldceo@bcherny@claudeai Very strange. I have not had a similar experience - I'm curious what the two prompts look like. The loops are kind of unbound, right? So direction matters a lot. Have you inspected your context to understand your context mix better?
it’s in gemini, just create it in ai studio. oh, that’s for your personal google one account. for workspace you need gemini business. no, not gemini advanced, that’s ai pro now. unless you need ai ultra. oh agents? you do that in spark actually. no, not gemini api managed agents, that’s different. for coding use jules. unless you mean the agentic ide, that’s antigravity. no, that’s the old antigravity, download the new one. actually gemini cli is being deprecated, use antigravity cli. no the flash model is smarter than the pro model. unless you need pro. if it’s video, use flow. no, flow uses veo. no, nano banana is images. actually that’s in gemini now. unless you’re in search, then it’s ai mode. no, research is notebooklm. anyway it’s all very simple.
Absolutely horrendous experience flying @cathaypacific - had a flight attendant pour an entire glass of wine on me, ruining my shirt, soaking my pants mid flight. Am told “talk to the ground crew” as I’m leaving the plane. Find the customer service agent at desk A in HKG and am told they can’t do anything for me because there’s “no evidence” despite me filling out an incident form mid-flight and the Airbus certainly being cleaned with a seat full of wine a few terminals down. I’m told they have no ability to contact the crew because they have “different systems” — An absolute joke. 15 hour flight to JFK with my clothing soaked with wine coming up. Zero compensation. Zero help. Basically told to kick rocks. Totally unacceptable
And that's the big unlock, I'm not just writing less tweets, I'm actively disengaged from Twitter discourse... and honestly? That's a powerful reframe. Let me break it down:
@garybasin@RhysSullivan You don’t even need background processes as part of the agent harness. Just tell the model to create a tmux session and reattach when it needs to check output
@OpenAI https://t.co/jS6IdlM4Z3 is a lighter weight version of this that is focused on gift giving. Super useful to find a great gift for a hard to shop for person
A claude slash command: https://t.co/yTae86ujTQ that provides context as to how you'd like to think about a set of documents. Maybe it summarizes the docs within a project folder, or provides guidelines for what you're focused on, the output gets passed down into other sub-agents, skills, prompts, etc.
Put all of your planning docs in an Obsidian Vault. Have a CLAUDE .md file in each project folder. Create planning commands at the vault level, tailor sub-agents at the project level. Build claude skills. Use claude code on your project docs and planning. #nomoreslop
We just shipped a new way to save and re-use system instructions in @GoogleAIStudio, making it easier to test, reproduce outputs, and experiment with Gemini : )
This has been a top ask for a while, glad we landed it!
I've pretty much moved on from Claude Code at this point. Aug-Sep was such a rollercoaster of availability and quality that I just can't trust it any more. I've been evaluating a ton of other solutions and made my workflow more flexible, the tooling lock-in just evaporated.